Skip to contentSkip to conversation

Large I/O operations result in slow performance and high iowait times

Bug 12309 Thomas Pilarski 2 participants 4 messages Mar 30, 2010 – Apr 4, 2010 2 messages not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]/

A message that is not in the archive (sent off-list or not yet received): [email protected]

#2 lore raw

Sorry for the late answer.

There is regression in desktop responsiveness from 2.6.32 to 2.6.33. I am working on a notebook with a full encrypted lvm partition. The throughput of the disk is about 70MB/s in the outer regions. There is another hard disk in an ultra bay slot.

While 2.6.32 is "usable", the 2.6.33 freezes for up to two minutes in a 30 seconds interval while tracker starts to index a new inserted hard disk with a lot of small files (openembedded build tree). This starts with a wrong configured tracker and the 2.6.32 kernel with freezes up to 10 seconds.

The indexed ultrabay dics is not encrypted and formated with ext4, only the indexes are writen on an ext3 partition in an ecrypted logical volume group, which also contains the system partition.

This problem is enormous increased, while the ultra bay disc is acceded with udma2 instead of udma6 (caused by another bug),

I can execute some tests, but you should tell me, which results you need to identify the problem.

Thomas Pilarski

Am Dienstag, den 16.03.2010, 23:57 +0000 schrieb
[email protected]:

Quoted 18 lines

http://bugzilla.kernel.org/show_bug.cgi?id=12309

--- Comment #420 from Andrew Morton <[email protected]> 2010-03-16 23:56:58 --- (In reply to comment #419)

Quoted 5 lines

I am currently using the linux kernel 2.6.33 and the the desktop responsiveness is awful on my machine compared to the 2.6.32.x kernel. It's even worse than I have even seen it before. The load avg is rising to >7 very quickly, while writing many small file to the filesystem. I can make some tests with my configuration, but a kernel developer should tell me which tests.

This isn't really the best place to bring this up. Please send a full description to [email protected]. cc myself, Ingo Molnar <[email protected]>, Peter Zijlstra <[email protected]>, Jens Axboe <[email protected]>. In that email, please identify what the system is doing at the time. Is it disk-related? CPU scheduler related? etc.

Thanks.

#3 lore raw
Quoted 2 lines · Thomas Pilarski

This problem is enormous increased, while the ultra bay disc is acceded with udma2 instead of udma6 (caused by another bug),

There is a thinkpad BIOS bug that could cause that: it will not signal a 80-wire cable when you hot-add an HD on a PATA ultrabay. Exists at least on all T4x, including the T43.

The fix is to whitelist these thinkpads for high-speed UDMA even on 40-wire cables. The workaround is to add a kernel command line parameter (I forget which).

Does that match your udma2 instead of udma6 problem?

Signature · 4 lines
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh
#4 lore raw
Quoted 6 lines
Quoted 2 lines

This problem is enormous increased, while the ultra bay disc is acceded with udma2 instead of udma6 (caused by another bug),

There is a thinkpad BIOS bug that could cause that: it will not signal a 80-wire cable when you hot-add an HD on a PATA ultrabay. Exists at least on all T4x, including the T43.

It's a SATA untrabay disk. There is a 40-wire message, but sometimes it works and sometimes not. I could not figure out, when it does and when not.

Quoted 5 lines

The fix is to whitelist these thinkpads for high-speed UDMA even on 40-wire cables. The workaround is to add a kernel command line parameter (I forget which).

Does that match your udma2 instead of udma6 problem?

Thanks. I will try it. But have to finish another project first.

#5 lore raw
Quoted 10 lines · Thomas Pilarski
Quoted 6 lines
Quoted 2 lines

This problem is enormous increased, while the ultra bay disc is acceded with udma2 instead of udma6 (caused by another bug),

There is a thinkpad BIOS bug that could cause that: it will not signal a 80-wire cable when you hot-add an HD on a PATA ultrabay. Exists at least on all T4x, including the T43.

It's a SATA untrabay disk. There is a 40-wire message, but sometimes it works and sometimes not. I could not figure out, when it does and when not.

Which ThinkPad model? And yes, if it is the BIOS 40-wire problem, it doesn't show up always. I keep forgetting exactly how to trigger it (I think it is hotplug device when you boot with the bay empty), but it is 100% reproducible.

Quoted 7 lines
Quoted 5 lines

The fix is to whitelist these thinkpads for high-speed UDMA even on 40-wire cables. The workaround is to add a kernel command line parameter (I forget which).

Does that match your udma2 instead of udma6 problem?

Thanks. I will try it. But have to finish another project first.

Just make sure to tell me if the ata 40-wire short cable override fixes your problem when you get around to trying it.

Signature · 4 lines
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh